Embed Size (px)
Citation preview
AN EXAMINATION OF FACTORS INFLUENCING THE SENTENCING OF CONVICTED FELONS IN ILLINOIS
Megan Alderden, Ph.D.Illinois Criminal Justice Information Authority Acting Executive Director
State of IllinoisJB Pritzker, Governor
An Examination of Factors Influencing the Sentencing of Convicted Felons in Illinois
Prepared by
David E. Olson, Ph.D.
&
Donald Stemen, Ph.D.
Loyola University Chicago
Department of Criminal Justice and Criminology
&
The Center for Criminal Justice Research, Policy and Practice
May 2019
This project was supported by Grant #16-DJ-BX-0083, awarded to the Illinois Criminal Justice Information Authority by
the U.S. Department of Justice Office of Justice Programs’ Bureau of Justice Assistance. Points of view or opinions
contained within this document are those of the authors and do not necessarily represent the official position or policies
of the U.S. Department of Justice.
Introduction
Between 1985 and 2000, the number of individuals convicted of felonies who were on probation,
in prison, or on parole in Illinois more than doubled, increasing from below 60,000 individuals to
more than 120,000 individuals (Figure 1).1 Since 2000, the number of individuals convicted of
felonies under the custody of Illinois’ justice system has hovered around 130,000 annually.
In addition to these changes in the overall number of people under correctional supervision in
Illinois, there has also been a change in the distribution of individuals within each component of the
correctional system. Between 1985 and 2000, roughly 50% of individuals convicted of felonies in
Illinois’ were under probation supervision (Figure 2). However, since 2000, this proportion has
dropped to roughly 40%, meaning that approximately 60% of individuals in Illinois’ criminal justice
system are currently in prison or have recently been released from prison onto parole. More
specifically, 37% of individuals under the supervision of Illinois’ correctional system in 2015 were
incarcerated in an Illinois prison, and an additional 22% were being supervised on mandatory
supervised release (MSR) following their release from prison.
1 Analyses by Loyola’s Center for Criminal Justice Research, Policy and Practice of published data from the Illinois Administrative Office of the Illinois Courts and the Illinois Department of Corrections.
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
19
85
19
86
19
87
19
88
19
89
19
90
19
91
19
92
19
93
19
94
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
20
09
20
10
20
11
20
12
20
13
20
14
20
15
Num
ber
of
convic
ted f
elo
ns
Figure 1. Number of individuals convicted of a felony under correctional supervision in Illinois, 1985-2015
The shift in the number and distribution of individuals under correctional supervision in Illinois
is partially due to a change in the use of prison in Illinois. Statewide, the proportion of individuals
sentenced to prison following conviction for a felony increased slightly between 1985 and 2009,
from roughly 42% of individuals receiving a prison sentence to 50% of individuals receiving a
prison sentence (Figure 3).2 During this period there was also a large increase in the number of
individuals processed through Illinois’ courts—jumping more than 100% between 1985 and 2000.
Taken together, the significant increase in the number of felony cases handled by the courts
combined with an increased likelihood of receiving a prison sentence following a felony conviction
caused a significant growth in the number of individuals under the custody of the Illinois
Department of Corrections-either in prison or MSR.
2 Analyses by Loyola’s Center for Criminal Justice Research, Policy and Practice of published data from the Illinois Administrative Office of the Illinois Courts.
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
19
85
19
86
19
87
19
88
19
89
19
90
19
91
19
92
19
93
19
94
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
20
09
20
10
20
11
20
12
20
13
20
14
20
15
Num
ber
of
convic
ted f
elo
ns
Figure 2. Number of individuals convicted of a felony under corrections supervision in Illinois, by status, 1985-2015
Probation Prison MSR "Parole"
However, when the proportion of individuals sentenced to prison is examined across each of
Illinois’ 102 counties, there is substantial variation in the rate of prison utilization (Figures 4 and 5).3
For example, DuPage County consistently has sent a smaller proportion of individuals convicted of a
felony to prison, relative to the state average, while Cook County consistently has sent a larger
proportion to prison. Winnebago County and Madison County, for example, experienced a steady
decrease in the proportion of individuals sentenced to prison, while Sangamon County experienced a
slight increase.
Figure 4. Percent of individuals convicted of a felony sentenced to prison in Illinois, by select counties, 1985-2015
3 Analyses by Loyola’s Center for Criminal Justice Research, Policy and Practice of published data from the Illinois Administrative Office of the Illinois Courts.
0%
20%
40%
60%
80%
100%
19
85
19
86
19
87
19
88
19
89
19
90
19
91
19
92
19
93
19
94
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
20
09
20
10
20
11
20
12
20
13
20
14
20
15
Perc
ent
of
convic
ted f
elo
ns
Figure 3. Percent of individuals convicted a felony sentenced to prison in Illinois, 1985-2015
By 2015, the differences in the use of prison across Illinois’ 102 counties was quite dramatic; in
some counties fewer than 20% of individuals convicted of a felony received a prison sentence, while
in other counties more than 60% of received a prison sentence (Figure 5).
* Each bar represents one county in Illinois. The bars are unlabeled to preserve the legibility of the graph.
Over the past 10 years, the state has put forth significant effort to understand and address the
high rate of incarceration in Illinois, including the development of the Adult Redeploy Illinois (ARI)
Winnebago
Boone DeKalb DuPage Grundy Jersey
Kendall Lake LaSalle Macon Madison
McLean Peoria Sangamon Will
2nd Circuit 4th Circuit 9th Circuit 20th Circuit Cook
0%
20%
40%
60%
80%
100%
Perc
ent
of
cases s
ente
nced t
o p
rison
Counties*
Figure 5. Percent of individuals convicted of a felony sentenced to prison in Illinois, by county, 2015
program, the establishment of the Illinois Sentencing Policy Advisory Council (SPAC), and, most
recently, the Illinois State Commission on Criminal Justice and Sentencing Reform. Despite these
significant efforts and advances, very little is known about what factors influence sentencing
practices in Illinois. While the research literature on sentencing has identified a number of
defendant- and case-level characteristics that influence sentencing, such as age, race, gender,
criminal history, current offense, etc., the influence of these characteristics on sentencing specifically
in Illinois has never been systematically or objectively examined. Although Illinois does not have all
of the data needed to do a comprehensive assessment of factors that influence sentencing practices,
there are sufficient data available that would allow the state to become more sophisticated in its
understanding of sentencing. Using what is currently available to examine sentencing practices will
likely lead to improved strategic planning, more effective use of fiscal and programmatic resources,
better individual outcomes and better protection of public safety. Further, determining the strengths
and weaknesses of the existent data can also aid the state in future efforts to enhance data availability
and quality.
Further, the community context and characteristics that influence sentencing decisions is a
dimension of sentencing research that has not received substantial attention in the published
literature, but could be advanced in Illinois. Increasingly, scholars are recognizing that, in addition to
defendant- and case-level characteristics, the characteristics of the community (county) likely
influences everything from sentencing decisions to recidivism. In other words, not only may the age,
gender, criminal history, and current offense of the defendant be important in explaining the
sentence imposed, but so too may be the extent and nature of crime in the community, the
availability of alternatives to prison, or the effectiveness of probation, to name just a few. In many
instances, decisions about whether and how to develop programs and strategies to reduce prison
utilization and improve public safety are made at the county-level. Thus, a more sophisticated
understanding of what county-level characteristics influence sentencing is critical to developing
meaningful and empirically-based responses.
This current research effort sought to utilize existing statewide data in Illinois to determine the
influence of defendant- and case-level characteristics on sentencing outcomes for those convicted of
a felony, and also to gauge the degree to which sentencing practices vary across Illinois counties
once these defendant- and case-level characteristics are taken into account.
Data and Methods
The current study utilizes data from the Illinois Criminal History Record Information (CHRI)
system. CHRI collects and maintains data for arrests occurring in Illinois, for the purpose of creating
individuals’ criminal history transcripts (commonly called “rap sheets”) Arrests are recorded in
CHRI when a law enforcement agency submits a defendant’s fingerprints, identifying information,
and charge information to the CHRI system as part of the arrest booking procedure. State’s attorneys
and Circuit Court clerks then report charging, disposition, and sentencing information to CHRI; thus,
CHRI records track individual arrest charges through final disposition and sentencing. CHRI data
contain detail on defendant characteristics, initial and final charge information, arrest dates,
disposition and sentencing dates, and final case dispositions and sentences. For a more detailed
description of CHRI, see the Illinois State Police’s CHRI User’s Manual.4
Working collaboratively with staff from the Illinois Criminal Justice Information Authority
(ICJIA), the Loyola research team obtained detailed, incident-level data on all arrests that occurred
4 https://www.isp.state.il.us/docs/5-336e.pdf.
in Illinois between 2012 and 2014 that could potentially result in a prison sentence. Detailed
information from CHRI was provided for every felony arrest, as well as any misdemeanor arrest that
could be elevated to a felony-level offense under specific circumstances (e.g., a second or
subsequent conviction that elevates a misdemeanor to a felony). The initial sample included
approximately 1,050,000 unique arrest charges associated with 450,000 unique arrest events and
approximately 373,000 unique defendants.5 Once unique individuals were identified from this initial
sample, the CHRI data were then used to determine all prior arrest events occurring for these
individuals prior to the study period (i.e., prior to 2012 through 2014). Since the final analyses
examine sentencing outcomes, the dataset was further limited only to individuals convicted of a
felony offense; this final sample included 115,442 unique defendants.
A multilevel strategy was used to estimate the independent effects of defendant, case, and county
characteristics on sentencing outcomes. The unit of analysis was the arrest event and analyses
tracked the sentence for each arrest event. Since arrest events often involve multiple charges, a
procedure was devised to categorize and track cases according to the “top charge” at each stage of
the process. To determine the top arrest charge, all arrest charges within a unique arrest event were
first ranked by offense severity according to the state’s three-part misdemeanor and six-part felony
classification system (Table 1). Arrest charges were then classified into five distinct offense types
(violent, illegal possession of weapons, property, drug-law violations, and “other”). The arrest
charge with the highest offense severity in a case was designated as the top arrest charge for analysis
purposes. When an event contained two arrest charges with the same offense severity but different
offense types, arrest charges were ranked according to offense type in the following way: violent
(most serious), weapons, property, drugs, other (least serious); when a case contained two arrest
charges with the same offense severity and the same offense type, charges were allowed to randomly
select as the top charge. To determine the top conviction charge, a similar strategy was used. All
charges resulting in a conviction were ranked according to offense severity and offense type as
above; the conviction charge with the highest offense severity in a case was designated as the top
conviction charge for analysis purposes.
5 Defendant identifying information was masked in the final dataset, through an algorithm that scrambled the defendant
ID; thus, it was not possible to identify individuals from the final data. Initial analyses of the data also indicated that
individual defendants were often associated with multiple arrest events being processed at the same time. In such
instances, the study treated separate arrest events that involved the same defendant as independent events.
Table 1. Penalties for Felony Crimes in Illinois
Category of Crime Imprisonment Term Probation Term Length of MSR (Post-prison supervision)
1st Degree Murder 20-60 years Not allowed 3 years
Class X Felony 6-30 years Not allowed 3 years
Class 1 Felony 4-15 years Up to 4 years 2 years
Class 2 Felony 3-7 years Up to 4 years 2 years
Class 3 Felony 2-5 years Up to 2 ½ years 1 year
Class 4 Felony 1-3 years Up to 2 ½ years 1 year
Class A Misdemeanor <=1 year Up to 2 years --
Class B Misdemeanor <=6 months Up to 2 years --
Class C Misdemeanor <=1 month Up to 2 years --
Since separate charges in the same arrest event can result in different sentences, a procedure was
also devised to categorize cases according to the “top sentence” at sentencing. To determine the top
sentence, all sentences associated with conviction charges within a unique arrest event were ranked
by degree of the seriousness of the sentence imposed in the following order: prison (most serious),
jail, probation or supervision, fines/restitution/public service (least serious). The most serious
sentence associated with a unique arrest event was designated as the “top sentence” for the entire
arrest event. Similarly, if more than one charge in the same arrest event included a prison sentence,
the longest prison sentence associated with the arrest event was use as the controlling prison
sentence length for the entire arrest event.
Dependent Variables
The current study examined sentencing outcomes across two dependent variables. The first
dependent variable, Prison Sentence, captures whether the defendant received a prison sentence
following a conviction for a felony (0=non-prison sentence imposed for all conviction charges; 1=
prison sentence imposed for at least 1 conviction charge). These analyses rely on the full dataset of
defendants convicted of a felony (N=115,442). Across this sample, 48% of those convicted of a
felony offense in Illinois during the sampling period were sentenced to prison.
The second dependent variable, Prison Sentence Length, captures the length of the prison
sentence imposed (continuous, in months) for those individuals sentenced to prison. These analyses
rely on a reduced data set including only defendants sentenced to prison (N=55,560). Across this
sample, the average prison sentence imposed was more than 45 months or roughly 3.8 years.
Source: 2014. “Penalties for Crimes in Illinois,” published online by the Illinois General Assembly Legislative Research
Unit. http://ilga.gov/commission/lru/2014PFC.pdf
Defendant-Level Variables
In addition to indicating which felony arrests and convictions resulted in a prison sentence, the
CHRI data also provides additional information on individual defendants, including limited
demographic characteristics and detailed information about the extent and nature of the defendant’s
prior criminal history.
Defendant characteristics include information on demographic attributes of defendants, namely,
sex (0=Female, 1=Male), race (0= White, 1=Black), and age (continuous, in years). We include a
dichotomous variable measuring race because of the exceptionally small populations of specific
minority racial groups in most counties in Illinois, outside of Cook County. Ethnicity is also poorly
captured in the CHRI data. Since 2015, agencies have had the option of reporting Hispanic ethnicity
as a category within the race field in CHRI. However, given the sample selected for the current
research (arrests occurring between 2012 and 2014), very few cases (0.2% of those convicted of a
felony) reported Hispanic as the racial category. Thus, all minority racial groups represented in the
CHRI data (Asian, Native American, etc.) are excluded from the analyses, and the analyses only
consider case outcomes and sentencing outcomes for White and Black defendants.6 We also created
a categorical variable capturing defendant age, capturing whether or not the defendant is a young
adult (0=defendant 25 years of age or older; 1=defendant under 25 years of age).
Several measures of defendant criminal history were also derived from the CHRI data. Criminal
history measures included whether a defendant had five or more prior arrests (0=less than 5 prior
arrests; 1=5 or more prior arrests), whether the defendant had a prior felony conviction (0=no prior
felony conviction, 1=at least 1 prior felony conviction), and whether the defendant had a prior prison
sentence imposed (0=no prior prison sentence imposed; 1=at least 1 prior prison sentence imposed).
To determine prior felony convictions and prior prison sentences, the ranking strategy described
above was used to determine top conviction charge and top sentence for prior arrest, conviction, and
sentencing events.
Case-Level Variables
In addition to sentencing and defendant information, the CHRI data also include additional
information on charges and case processing. Charge characteristics include the number of conviction
charges (continuous) and the severity of the most serious conviction charge, coded as an ordinal
variable with nine categories (0=Class C misdemeanor [least serious] through 8=Murder [most
serious]). Because cases may include convictions for charges that carry mandatory prison sentences,
a dummy variable is used to capture whether a final conviction charge is non-probationable (0=case
does not include non-probationable conviction charge; 1=case includes non-probationable conviction
charge). The type of offense is also measured with dummy variables for violent, weapons, property,
drugs, and other offenses.
Some research has found that the longer an individual is held in pre-trial detention, the greater
the likelihood of both conviction and a prison sentence following conviction. However, information
about pre-trial jail custody is not reported to CHRI. Although CHRI does not capture a defendant’s
pretrial status, a proxy measure was created that captures whether a convicted defendant received
any credit for time served (0=did not receive credit for time served; 1=received credit for time
served).7 Analyses also include a measure capturing whether the arrest event included an arrest for a
6 Defendants of other racial/ethnicity groups accounted for just 1.6% of all defendants in the CHRI data. 7 CHRI only captures credit for time served when an individual is sentenced; in such instances, CHRI reports how many
days of credit for time served the individual received as part of the imposed sentence. If an individual receives a sentence
weapons offense (0=arrest event does not include arrest for weapons offense; 1= arrest event include
arrest for weapons offense). Research also has found that sentences are influenced by the mode of
conviction (guilty plea versus a trial), with convictions resulting from a trial having a higher
likelihood of receiving a prison sentence and a longer prison sentence than those resulting from a
guilty plea. However, CHRI does not include information on whether a conviction is obtained
through a guilty plea or trial; thus, a proxy for guilty pleas was created by capturing whether there
was a reduction in the number of charges from arrest to conviction (0=no reduction in charges from
arrest to conviction; 1=reduction in charges from arrest to conviction) and if there was a reduction in
the severity of the charges from arrest to conviction (0=severity of charges reduced from arrest to
sentencing, 1=no change in severity of charges from arrest to sentencing, and 2=severity of charges
increased from arrest to sentencing).
County-Level Variables
In addition to examining the influence of defendant- and case-level characteristics, we also
include several county-level characteristics. County characteristics include the index crime rate
(continuous) and the index crime “clearance rate” (continuous, calculated by taking arrests divided
by offenses reported) as reported to the Illinois State Police through the Illinois Uniform Crime
Report for 2012. Several county measures are derived from 2012 the Annual Report of the
Administrative Office of the Illinois Courts, including the felony filing rate (number of felony filings
per capita, continuous), the guilty plea rate (the percent of felony convictions resulting from a guilty
plea, continuous), and the prison utilization rate (the percent of felony convictions resulting in a
sentence to prison, continuous). We also included several measures of county population
characteristics derived from the 2010 U.S. Census, including the percent of the county population
that is white (continuous), the median household income (continuous), and population density
(continuous). We included a measure of civic engagement, which captures the percent of eligible
voters voting in the 2010 general election (continuous), and a measure of political liberalism, which
captures the percent of votes cast for the democratic gubernatorial candidate in the 2010 election
(continuous). Finally, some argue that counties without a jail may be more likely to use prison as a
sentencing option since they lack a local option to incarcerate individuals following a conviction;
thus, we include a variable capturing whether the county has a jail located within its boundaries
(0=county does not have a jail within its borders; 1=county has a jail within its borders).
Analytical Strategy
The impact of defendant, charge, and case characteristics on case outcomes and sentences were
analyzed using standard statistical procedures to examine categorical data in multivariate settings.
Specifically, our baseline estimations rely on a series of logistic regression models and ordinary least
squares models to estimate the effect of these factors on the case outcome measures described above.
Because of the nested nature of the data (cases nested within counties) the analyses also relied on
hierarchical generalized linear modeling (HGLM) procedures designed to account for the nested
nature of multilevel data. Cases processed in the same county are likely to have certain similarities;
as a result, residual errors are likely to be correlated within counties, violating fundamental error
of probation plus credit for time served in jail, it would be fair to assume the defendant spent some time in pre-trial
detention. Thus, if this information is accurately and consistently included as part of the sentencing information reported
through CHRI, it may be an accurate measure of the amount of pre-trial detention time spent in jail. However, the
accuracy and quality of this information has never been audited or evaluated in Illinois. Moreover, since defendants
received time served only if convicted, this measure was used only in analyses examining sentences to prison and prison
sentence length.
assumptions of regression techniques and resulting in misestimated standard errors. HGLM resolves
this problem by incorporating into the statistical model a unique random effect for each county
(Raudenbush & Bryk, 2002). HGLM also allows for the modeling of the heterogeneity of regression
coefficients that can occur when relationships between individual characteristics and outcomes vary
across aggregate units. For example, the effect of being a minority defendant or being convicted of a
drug offense may differ across counties. Moreover, combining these county-level data with the
defendant- and case-level data allows the examination of how individual- or county-level factors
influence sentencing decisions, and provides an opportunity to identify specific county
characteristics that influence sentencing decisions that can be addressed through policy. For
example, county-level factors such as the capacity of drug treatment within a county may influence
the county’s utilization of prison - something that could be addressed through targeted allocation of
treatment resources.
A two-level hierarchy represents the current data, with individual cases nested within counties.
All continuous variables are centered on their grand means and results reported are based on unit-
specific models using robust standard errors (Raudenbush & Bryk, 2002). All models are estimated
using Stata 14.
The analyses begin by estimating models with defendant- and case-level predictors with simple
dummy variables for counties. The HLM models then estimate unconditional models without any
predictors; these models produce estimates of the relative amount of variation that occurs at the case-
and county-levels of analysis, providing insights into the relative importance of the county in case
sentencing outcomes. Case-level predictors are then added to the models to assess the degree to
which county-level variations are accounted for by compositional differences in cases. This provides
information on the extent to which case-level factors vary significantly across counties. Two-level
hierarchical models are then estimated to investigate the direct effects of specific county-level
characteristics on outcomes.
Results
Descriptive Statistics
Table 2a provides descriptive statistics for the final samples used in the analyses of whether or
not a defendant convicted of a felony was sentenced to prison, and of those sentenced to prison, the
length of the sentence imposed. As Table 2a indicates, just over 48% of those convicted of a felony
in Illinois during the sampling period were sentenced to prison,8 with an average sentence length of
3.8 years.
Column 1 in Table 2a includes descriptive statistics for all individuals convicted of a felony in
Illinois during the study period. Nearly 84% of individuals convicted of a felony were male, 50%
were non-white, and 34% were under the age of 25 (i.e., “emerging adults”). In terms of the prior
criminal history, 62% of those convicted of a felony had five or more prior arrests, 56% had at least
one prior conviction for a felony offense, and 36% had a prior prison sentence.
Within the sample of individuals convicted of a felony, 43.4% were convicted of a Class 4
felony, which is the least serious felony offense class; a very small percent of individuals were
8 This rate is consistent with that seen in published, aggregate data reported by Clerks of the Circuit Courts across Illinois
to the Administrative Office of the Illinois Courts (AOIC). Specifically, statewide, 44% of felons convicted between
2012 and 2015 in Illinois were sentenced to prison based on the published AOIC data.
convicted of Murder (0.2%) or a Class X felony (5.2%). Roughly 10% of individuals convicted of a
felony were convicted of a non-probationable offense, meaning that the conviction required the
imposition of a prison sentence. In terms of the nature of the current conviction offense, the majority
of individuals were convicted of a non-violent offense – almost one-third (32.8%) were convicted of
a drug-law violation and 28% were convicted of a property crime; just 12% were convicted of a
violent crime and another 6% were convicted of a weapon offense (primarily the illegal possession
of a firearm).9 Just 7.5% of cases involved an arrest for a weapons offense.
Finally, nearly 60% of individuals convicted of a felony received credit for time served,
indicating that a high percentage of individuals likely were detained at some point pretrial. In
addition, roughly 41% of individuals received a reduction in the number of charges from arrest to
conviction.
Column 2 in Table 2a includes descriptive statistics for all individuals sentenced to prison in
Illinois during the study period. This population is slightly different than the population of
individuals convicted of a felony. Over 90% of individuals sentenced to prison were male, 60% were
non-white, and 32% were under the age of 25. Individuals sentenced to prison had more serious
criminal histories than the overall sample of convicted individuals; 78% of those sentenced to prison
had five or more prior arrests, 74% had at least one prior conviction for a felony offense, and 56%
had a prior prison sentence.
Individuals sentenced to prison were similar to those convicted of a felony in terms of offense
severity and offense type; 38% were convicted of a Class 4 felony, just 0.4% were convicted of
Murder, and 8.5% were convicted of a Class X felony. Similar to those convicted of a felony, 31%
of individuals sentenced to prison were convicted of a drug-law violation, 28% were convicted of a
property crime, and just 23% were convicted of a violent offense or a weapon offense. Overall, 16%
of individuals sentenced to prison were convicted of a non-probationable offense; in other words,
nearly 84% of individuals sentenced to prison could have been sentenced to probation.
Finally, nearly 73% of individuals sentenced to prison received credit for time served, indicating
that a high percentage of individuals likely were detained at some point pretrial. In addition, roughly
42% of individuals sentenced to prison received a reduction in the number of charges from arrest to
conviction.
9 Crimes committed with a firearm, such as armed robbery or aggravated battery with a firearm, were classified as
violent offenses, not weapon offenses.
Table 2a. Descriptive Statistics for Defendant- and Case-level
Variables
Convicted
Sentence to
Prison
N 115,442 55,560
Dependent Variables Mean/% Mean/%
Prison Sentence 48.1% --
Prison Sentence Length (years) -- 3.8
Individual-level Variables
Defendant Male 83.8% 90.1%
Defendant Non-White 50.5% 60.2%
Defendant Young Adult 34.0% 31.9%
5 or more prior arrests 61.8% 78.0%
Prior Felony Conviction 56.0% 74.3%
Prior Prison Sentence 36.0% 55.9%
Number of Charges 1.1 1.1
Offense Severity
Murder 0.2% 0.4%
Class X 5.2% 8.5%
Class 1 10.5% 13.6%
Class 2 20.6% 23.1%
Class 3 19.3% 16.2%
Class 4 43.4% 38.1%
Non-probationable Conviction 10.0% 16.4%
Conviction Offense
Violent 12.0% 13.8%
Weapons 6.1% 9.4%
Property 28.3% 27.7%
Drugs 32.8% 30.5%
Other 20.8% 18.6%
Credit for Time Served 58.9% 73.2%
Arrest for Weapons Offense 7.5% 11.1%
Reduction in # of Charges 41.4% 41.9%
Time from Arrest to Disposition (months) 7.3 7.7
Table 2b provides descriptive statistics for the county-level variables. As Table 2b indicates,
there is significant variation across counties in both the dependent variables and county-level
variables. While 48% of those convicted of a felony in Illinois overall during the sampling period
were sentenced to prison,10 the percentage of those sentenced to prison by county ranged from a
minimum of 21% in one county to over 63% in another county. Similarly, while the average prison
sentence length was 3.8 years across the state, this ranged from an average sentence of just 1 year in
one county to over 6 years in another.
Table 2b also indicates that Illinois counties are quite diverse on a number of important factors.
Counties have markedly different crime rates. Although the average index crime rate was 3,914
offenses per 100,000 populations, some counties had crime rates as low as 675 while others had
crime rates as high as 10,000. Given these differences in crime rates, counties also experienced
markedly different caseloads in the courts, with some counties handling relatively few cases (511
cases per capita) and other counties having very high caseloads (3,274 cases per capita). Counties
were similarly diverse on measures of demographics, income, voter participation, and political
liberalism.
Table 2b. Descriptive Statistics for County-level Variables
Dependent Variables Mean/% Min
Max
Prison Sentence 48.1% 21% 63%
Prison Sentence Length (years) 3.8 1.0 6.2
County-level Variables (N=102)
Index Crime Rate 3,914.4 675.72 10,529.03
Index Crime “Clearance Rate” 26.3% 5.0% 100.0%
Felony Filing Rate 1,567.9 511.45 3,274.49
Guilty Plea Rate 95.6% 51.0% 100%
Prison Utilization Rate 41.8% 22.0% 71.0%
Percent Population Non-White 11.5% 2.0% 56.0%
Median Household Income $49,765 $25,495 $83,844
Population Density 196.2 12.1 5,495.1
Voter Turnout 50.3% 37.2% 68.3%
Percent Voting for Dem. Gov. 34.2% 17.0% 69.0%
Jail in County 89.2% -- --
Bivariate Analyses
In order to examine the relationship between the independent variables and the initial sentencing
outcome (whether or not the individual convicted of a felony was sentenced to prison), a two-step
process was employed. The first step involved bivariate analyses comparing each individual
independent variable to the dependent variable (whether or not sentenced to prison). The second step
involved multivariate analyses, whereby all of the independent variables were examined
10 This rate is consistent with that seen in published, aggregate data reported by Clerks of the Circuit Courts across
Illinois to the Administrative Office of the Illinois Courts (AOIC). Specifically, statewide, 44% of felons convicted
between 2012 and 2015 in Illinois were sentenced to prison based on the published AOIC data.
simultaneously to determine their influence on the imposition of a prison sentence after the other
variables included in the analyses were statistically controlled.
Table 3 in the Appendix presents the bi-variate analyses, revealing a relatively weak, albeit
statistically significant, relationship between age and whether or not a prison sentence was imposed.
Specifically, the older the individual, the greater their likelihood of receiving a prison sentence. For
example, 45% of individuals between 18 and 24 convicted of a felony were sentenced to prison,
compared to 50% of those in the older age groups. Again, this is only looking at age and not taking
into consideration other factors, such as criminal history or the nature of the current conviction
offense. There was also a relationship between race and gender and the sentence imposed. Black
defendants convicted of a felony were more likely than white defendants to be sentenced to prison
(57% versus 39%, respectively), and men convicted of a felony were more likely than women to be
sentenced to prison (52% versus 29%, respectively).
As would be expected, those convicted of more serious felony class offenses were more likely to
be sentenced to prison. Among those convicted of a Class 4 felony, the least serious felony offense,
42% were sentenced to prison, compared to 54% of those convicted of a Class 2 felony and 62% of
those convicted of a Class 1 felony. Further, more than 70% of those convicted of weapon offenses
were sentenced to prison (again, these offenses are primarily the illegal possession of firearms and
do not include the use of a gun in the commission of a violent crime), and 55% of those convicted of
a violent crime were sentenced to prison.
Finally, the more extensive a person’s criminal history record the greater the chances they were
sentenced to prison. For example, 64% of those with a prior felony conviction were sentenced to
prison for the current felony offense, compared to 28% of those without a prior felony conviction.
Similarly, 75% of those who had previously been sentenced to prison received a prison sentence for
their current felony conviction, whereas 33% of those without a prior prison sentence received a
prison sentence for the felony conviction being studied.
Multivariate Analyses of Likelihood of a Prison Sentence
When multivariate analyses were performed, a number of the patterns found in the bivariate
analyses remained the same, but some changed (Table 4). For example, examining all individuals
convicted of felony in Illinois, the analyses showed that those between the ages of 18 and 24 were
more likely to be sentenced to prison than older felons after statistically controlling for all of the
other defendant- and case-characteristics. Further, after statistically controlling for all of the other
characteristics, the defendant’s race appeared to have a statistically significant, independent effect on
whether or not a prison sentence was imposed. Specifically, Black defendants were slightly more
likely to be sentenced to prison than White defendants even after taking into account things like
criminal history, gender, age, and the nature of the current offense. However, the magnitude of the
effect of race was substantively smaller than the effect of all other variables included in the analyses
(age, gender, criminal history measures, and characteristics of the current conviction offense). Black
defendants were found to be 8% more likely to be sentenced to prison than White defendants after
the effects of all of these other characteristics were statistically controlled.
Further, the influence that race had on sentencing appeared to be sensitive to whether or not the
analyses were run for Illinois as a whole or if Cook County was examined separately from the rest of
the state. When only cases in Cook County were examined, race was independently related to the
sentence imposed, with Black defendants being more likely to be sentenced to prison than White
defendants after controlling for other characteristics. However, when all cases outside of Cook
County were examined, the influence of race was the opposite of that seen in Cook County.
Specifically, of those sentenced in the combined area of Illinois outside of Cook County, Black
defendants were less likely to be sentenced to prison than White defendants after statistically
controlling for the other characteristics included in the analyses. In both sets of analyses (Cook
County and Illinois outside of Cook County), race still played a relatively small role relative to the
influence of the other variables included in the analyses.
Most of the other patterns found in the bivariate analyses were consistent in the multivariate
analyses. Even after statistically controlling for the influence of all of the other characteristics of the
defendant and the current offense, men convicted of a felony were more likely to be sentenced to
prison than women, and the more extensive the criminal history the greater the likelihood of being
sentenced to prison for the current felony conviction. The strongest predictor of being sentenced to
prison was if the defendant had previously been sentenced to prison.
The proxy that was developed to gauge whether or not the defendant had been held pre-trial (the
inclusion of credit for time served as part of the sentence imposed) was the second most influential
characteristic affecting the imposition a prison sentence. Specifically, those who received credit for
time served as part of their sentence were more likely to be sentenced to prison than were those who
did not receive credit for time served. On the other hand, the proxy used to measure plea bargaining
(reduction in the number of charges from arrest to conviction) was not statistically related to whether
or not a prison sentence was imposed after controlling for the other defendant- and case-
characteristics.11 Similarly, the number of days between the arrest and case disposition was also not
correlated with whether or not a prison sentence was imposed after the other variables were
simultaneously included in the analyses.12
11 Although the reduction in the number of charges between arrest and conviction was not statistically associated with the
imposition of a prison sentence when all cases were examined at the same time, when the analyses were performed
separately for Cook County versus the rest of Illinois, different patterns were evident. In Cook County, a reduction in the
number of charges was independently associated with a reduced likelihood of a prison sentence being imposed, whereas
in the combined area of Illinois outside of Cook County, a reduction in the number of charges increased the likelihood of
a prison sentence being imposed. 12 Again, although the time between arrest and case disposition was not statistically associated with the imposition of a
prison sentence when all cases were examined at the same time, when the analyses were performed separately for Cook
County versus the rest of Illinois, different patterns were evident. In Cook County, the longer the case took to be
resolved, the higher the likelihood of a prison sentence being imposed, whereas in the combined area of Illinois outside
of Cook County, the longer the case took to be resolved, the lower the likelihood of a prison sentence being imposed.
Table 4. Logistic Regression Predicting Likelihood of a Prison Sentence
B Odds S.E. Wald Sig.
Defendant Male .491 1.634 .020 584.167 .000
Defendant Non-White .079 1.082 .015 27.788 .000
Defendant Young Adult .188 1.206 .016 130.706 .000
5 or more prior arrests .513 1.670 .018 808.674 .000
Prior Felony Conviction .587 1.799 .020 879.406 .000
Prior Prison Sentence 1.242 3.461 .019 4204.551 .000
Number of Charges .275 1.317 .018 236.890 .000
Offense Severity .263 1.300 .007 1391.906 .000
Non-probationable Conviction .655 1.925 .031 439.540 .000
Violent Offense (reference) -- -- -- 422.634 .000
Weapon Offense .586 1.796 .063 86.403 .000
Property Offense -.094 .910 .025 14.597 .000
Drug Offense -.252 .777 .024 107.560 .000
Other Offense -.026 .975 .027 .911 .340
Missing Offense Type -.585 .557 .052 128.307 .000
Credit for Time Served .915 2.496 .015 3928.759 .000
Arrest for Weapons Offense .346 1.413 .054 40.310 .000
Reduction in # of Charges .003 1.003 .015 .047 .828
Time from Arrest to Disposition .001 1.001 .001 .912 .340
Constant -3.418 .033 .039 7590.046 .000
Analyses were also performed to determine the degree to which characteristics of the county of
conviction influenced the imposition of a prison sentence. Using hierarchical linear modeling, both
the defendant- and case-level characteristics and county-level characteristics were included in the
model, including the county Index crime rate, Index crime “clearance rate,” median household
income, percent of the population non-white, percent of the population voting (to measure civic
engagement), and the percent of the population voting for Democratic for governor. The analysis
first estimated an unconditional model without any predictors. This model produced estimates of the
relative amount of variation that occurs at the case- and county-levels of analysis, providing insights
into the relative importance of the county in sentencing decisions. Two-level hierarchical models
were then estimated to investigate the direct effects of specific county-level factors on outcomes.
Initial analyses indicated that there was significant variance in the likelihood of a prison
sentence across counties; overall, counties explained roughly 16% of the variance in the likelihood
of a prison sentence. Table 5 presents the results of the full HLM analyses. Although the results for
the initial null models showed that case-level data were significantly clustered within counties, the
county-level variables used in this study had no impact on the whether or not a convicted felon was
sentenced to prison; in other words, although counties explained 16% of the variance in the
likelihood of a prison sentence, this variance was not explained by differences in counties’ crime
rates, guilty plea rates, demographics, etc. Given the lack of significant impact on outcomes, Table 5
presents the defendant- and case-level effects on the likelihood of prison sentence after controlling
for the county-level effects on outcomes and a limited number of county-level effects.13
As Table 5 indicates, none of the county-level variables included in the analyses were
significantly related to the likelihood of a prison sentence; in other words, although much of the
variance in the imposition of a prison sentence can explained by the counties, variation across
counties cannot be explained by crime rates, population demographics, household income, or
democratic voting patterns. Table 5 also indicates that defendant- and case-level effects found
significant in the initial logistic models remain significant after controlling for county of conviction.
After controlling for county-level factors, being male, non-white, and younger increased a
defendant’s likelihood of a prison sentence after conviction. Defendants with more serious criminal
histories, convicted of more serious offenses, and held pretrial were also more likely to receive a
prison sentence. The only difference between the original logistic models and the HLM models was
the effect of reductions in the number of charges; after controlling for county-level factors,
defendants receiving a reduction in the number of charges at sentencing were more likely to receive
a prison sentence.
13 Given the limited number of level-2 observations in the analyses (102 counties), we included just four county-level
variables in each HLM model estimate; this is a conservative rule of thumb on the necessary number of cases necessary
for each variable included in the analyses. We ran several models including different combinations of county-level
variables; however, the HLM analyses revealed no significant effects of the any county-level variables. As such, we do
not report all of the analyses here.
Table 5. HLM Model Predicting Likelihood of a Prison Sentence
Odds S.E. Sig.
Null Model
Intercept 0.68 0.03 .000
Defendant Male 1.64 .033 .000
Defendant Non-White 1.05 .016 .001
Defendant Young Adult 1.21 .020 .000
5 or more prior arrests 1.67 .031 .000
Prior Felony Conviction 1.83 .036 .000
Prior Prison Sentence 3.43 .066 .000
Number of Charges 1.26 .018 .000
Offense Severity 1.28 .009 .000
Non-probationable Conviction 1.96 .062 .000
Violent Offense (reference) -- -- .000
Weapon Offense 1.55 .098 .000
Property Offense 0.88 .022 .000
Drug Offense 0.71 .017 .000
Other Offense 0.91 .025 .002
Credit for Time Served 2.87 .047 .000
Arrest for Weapons Offense 1.52 .083 .000
Reduction in # of Charges 1.10 .017 .000
Time from Arrest to Disposition 1.00 .000 .895
Constant 0.04 .003 .000
Index Crime Rate 0.99 .000 .749
% of the Population Nonwhite 0.77 .629 .757
Median Household Income 0.99 .667 .980
% of Population Voting Democratic 0.57 .495 .519
Although the specific county-level factors included in the HLM models were not significant, the
analyses indicate that there is considerable variation across counties in the likelihood of a prison
even after taking into consideration the influence of the defendant’s demographic, criminal history,
and current conviction offense characteristics. To illustrate this, a series of regression models were
run using all of the individual-level measures described above, plus a variable indicating whether or
not the conviction occurred in a specific county. For illustrative purposes, separate models were run
to examine the degree to which the odds of a prison sentence were higher or lower in each of the 30
counties that sent the largest number of people to prison in 2015 (Figure 6). These analyses revealed
that the odds of a defendant being sentenced to prison was higher in some counties even after taking
into account the influence of the defendant’s demographic, criminal history and current conviction
offense characteristics. For example, the odds of being sentenced to prison was 28% higher in Cook
County than elsewhere in Illinois after statistically controlling for the defendant and case
characteristics described above. On the other hand, in Winnebago County (the second largest county
in Illinois in terms of the volume of admissions to prison) the odds of being sentenced to prison was
almost 50% lower than the rest of Illinois after controlling for the defendant and case characteristics.
Multivariate Analyses of Prison Sentence Length
Multivariate analyses were also performed to determine the influence of the defendant and case
characteristics on the length of the prison sentence imposed for those who were sentenced to prison
(Table 6). For a number of the defendant and case characteristics, the direction of the relationship
changed when examining sentence length. For example, while emerging adults (those between 18
and 24) were more likely to be sentenced to prison, they received shorter prison sentences than older
adults. Similarly, those who received credit for time served in jail as part of their sentence were more
likely to be sentenced to prison, but tended to receive shorter sentences than those that did not
receive this credit.
The other variables included either had the same type of influence on sentence length as they did
on whether or not the defendant was sentenced to prison, or they had no effect on sentence length
but did influence the decision regarding a prison sentence. For example, men were more likely than
women to be sentenced to prison, and also received longer prison sentences. Similarly, those with a
prior prison sentence were more likely to be sentenced to prison after controlling for the other
characteristics; they also received longer prison sentences. As would be expected, those sentenced
for more serious felonies, for non-probationable offenses, and for violent offenses also received
longer prison sentences. Race, prior felony convictions, the total number of guilty charges, and
reductions in the number of charges did not have a statistically independent effect on prison sentence
length.
JeffersonMcLean
ColesKane
SangamonKnox
McHenryWinnebagoRock Island
LakeChampaign
St. ClairJackson
MaconMadisonVermillion
BooneDeKalbKankakeePeoriaStephensonTazewell
DuPageWill
CookLasalle
WhitesideHenry
MarionAdams
Saline
-100% -50% 0% 50% 100% 150% 200%
Percent change in odds of prison sentence
Table 6. County Effects on Odds of a Prison Sentence
Table 6. HLM Model Predicting Likelihood of a Prison Sentence
Model
Unstandardized
Coefficients
Standardized
Coefficients
t Sig. B Std. Error Beta
1 (Constant) -21.932 1.446 -15.169 .000
Defendant Male 3.739 .923 .016 4.051 .000
Defendant Non-White -1.075 .585 -.008 -1.837 .066
Defendant Young Adult -4.152 .628 -.028 -6.608 .000
5 or more prior arrests -1.610 .791 -.010 -2.034 .042
Prior Felony Conviction -.974 .879 -.006 -1.108 .268
Prior Prison Sentence 7.489 .734 .053 10.202 .000
Number of Charges .018 .529 .000 .034 .973
Offense Severity 16.642 .257 .326 64.809 .000
Non-probationable Conviction 21.383 .959 .114 22.298 .000
Credit for Time Served -5.356 .616 -.034 -8.689 .000
Arrest for Weapons Offense 2.500 1.754 .011 1.426 .154
Reduction in # of Charges .709 .565 .005 1.254 .210
Time from Arrest to Disposition .810 .037 .089 21.900 .000
Violent offense 23.219 .836 .115 27.761 .000
Weapon offense -15.167 1.893 -.064 -8.011 .000
Analyses were also performed to determine the degree to which characteristics of the county of
conviction influenced the length of the prison sentence imposed. Using hierarchical linear modeling,
both the defendant- and case-level characteristics and county-level characteristics were included in
the model. Again, these models produced estimates of the relative amount of variation that occurs at
the case- and county-levels of analysis, providing insights into the relative importance of the county
in the lengths of prison sentences.
Initial analyses indicated that there was significant variance in the lengths of prison sentences across
counties; overall, counties explained roughly 31% of the variance in the length of a prison sentence.
Table 7 presents the results of the full HLM analyses. Although the results for the initial null models
showed that case-level data were significantly clustered within counties, the county-level variables
used in this study had no impact on the length of a prison sentence. In other words, although counties
explained 31% of the variance in the length of a prison sentence, this variance was not explained by
differences in counties’ crime rates, guilty plea rates, demographics, or other county characteristics.
Given the lack of significant impact on outcomes, Table 7 presents the defendant- and case-level
effects on the length of a prison sentence after controlling for the county-level effects on outcomes
and a limited number of county-level effects.14
14 Given the limited number of level-2 observations in the analyses (102 counties), we included just four county-level
variables in each HLM model estimate; this is a conservative rule of thumb on the necessary number of cases necessary
for each variable included in the analyses. We ran several models including different combinations of county-level
variables; however, the HLM analyses revealed no significant effects of the any county-level variables. As such, we do
not report all of the analyses here.
Table 7. HLM Model Predicting Prison Sentence Length
B S.E. Sig.
Null Model
Intercept 5.58 .581 .000
Defendant Male 3.60 .427 .000
Defendant Non-White 0.69 .349 .048
Defendant Young Adult -1.26 .343 .000
5 or more prior arrests 1.59 .400 .000
Prior Felony Conviction 2.02 .441 .000
Prior Prison Sentence 10.15 .422 .000
Number of Charges 0.03 .436 .895
Offense Severity 10.90 .147 .000
Non-probationable Conviction 27.31 .617 .000
Violent Offense (reference) -- -- .000
Weapon Offense -25.14 1.26 .000
Property Offense -15.66 .533 .000
Drug Offense -15.08 .522 .000
Other Offense -13.09 .587 .000
Credit for Time Served -5.85 .346 .000
Arrest for Weapons Offense 5.02 1.15 .000
Reduction in # of Charges 0.95 .334 .004
Time from Arrest to Disposition 0.01 .000 .000
Constant -3.39 .999 .001
Index Crime Rate 0.00 .003 .340
% of the Population Nonwhite 1.65 8.11 .838
Median Household Income -0.00 .000 .176
% of Population Voting Democratic 0.061 8.41 .994
As Table 7 indicates, several defendant- and case-level effects found significant in the initial
regression models changed direction after controlling for county of conviction. For example, after
controlling for county-level factors, being nonwhite was associated with a longer imposed prison
sentence, although the effect is small and nearly non-significant. Similarly, although not significant
in the initial models, prior arrests and prior felony convictions were associated with longer prison
sentences after controlling for county of conviction. Finally, after controlling for county-level
factors, receiving a reduction in the number of charges at sentencing was also associated with a
slight increase in the prison sentence imposed.
Conclusions/Limitations
The current research project was the first known use of CHRI data to conduct a systematic
examination sentencing practices across Illinois. Although there remain a number of limitations with
using CHRI to fully understand the factors that explain whether or not those convicted of a felony
are sentenced to prison, these analyses do contribute some important findings for criminal justice
practice, policy and research in the state. First, the research found that there are a number of specific
defendant-level characteristics that consistently explained whether or not a prison sentence was
imposed following conviction of a felony, regardless of the nature of the conviction offense,
including age (younger individuals were more likely to be sentenced to prison than older individuals)
and gender (men convicted of a felony were more likely to go to prison than women). However, a
person’s prior criminal history played the strongest role in explaining variations in the imposition of
a prison sentence. The research highlights the significant implications of a sentence to prison for any
future felony charges an individual may have filed against them; all things being equal, individuals
previously sentenced to prison were much more likely to be sentenced to prison again. Conversely,
the research also found that those previously sentenced to probation were also more likely to be
sentenced to prison. Importantly, prior arrests, regardless of prior convictions or sentences to prison
also increased the likelihood of a prison sentence.
Given the detail of the information available in the CHRI data, particularly the fact that the data
are individual-level and county-specific, the types of analyses presented in this report can easily be
replicated across individual counties in Illinois to examine the degree to which the findings in these
statewide analyses are evident across individual counties. As illustrated in these analyses, the
influence (both in terms of direction and strength) of specific defendant and current offense
characteristics on whether or not a prison sentence was imposed varies when Cook County was
examined alone, or when Illinois outside of Cook County was examined separately. Indeed, in
research being performed for the ICJIA/Loyola University Chicago Criminal Justice Coordinating
Council pilot project, analyses using multivariate models similar to those developed and presented in
this report were performed for Winnebago, McHenry, Lake, McLean and St. Clair counties . These
county-specific analyses found that there were some defendant and offense characteristics that were
statistically associated with the imposition of prison sentences in some counties but not in others,
reflecting the unique sentencing environments of individual counties. Despite the efforts in the
current analyses to determine what specific county characteristics might explain these differences,
the county-level measures considered here were not found to explain any of the variation in the use
of prison across the counties. Significantly, defendants convicted in counties with high crimes rates,
lower clearance rates, higher case filing rates, or different voting rates of voter participation or
voting patterns were not sentenced differently due to these variations in county context. Still, the
analyses presented here highlight that even after accounting for differences between the
characteristics of the defendants or the nature of the felony cases handled, there remains
considerable variation in the use of prison from county to county.
In terms of the limitations, although the CHRI data provide a great deal of detail regarding
defendant criminal history and the outcomes of arrests (i.e., court filings, case dispositions, and
sentencing), CHRI does not contain some specific information that has been found in the literature to
explain sentencing practices. For example, some research has found variation in the sentences
imposed between those defendants represented by private counsel versus a public defender, those
held in detention pre-trial, and those cases disposed of through a trial rather than a guilty plea; CHRI
does not contain any of this information. This gap limits the ability to fully understand the factors
that influence sentencing practices in the state as a whole, and may also impact the direction and
magnitude of effect of those variables that were included in the analyses. While it is possible to
combine the CHRI data with these data from individual counties by requesting and accessing these
data through local data collection systems, to do so statewide would be very cumbersome given that
there are 102 counties in the state, most with different data systems. And, as illustrated above, it
cannot be assumed that factors that explain sentencing in one county are generalizable to the state as
a whole.
Despite these limitations, the methodology and technique used to perform these analyses could
be replicated using a pre-post design to assess the impact of specific interventions/programs the state
has implemented to reduce reliance on prison. With an intervention such as ARI, for example, the
goal of the program is to divert specific targeted populations from prison, such as felony drug
possession offenders, retail theft cases, etc. We may observe that the proportion of convicted felons
sentenced to prison went down after the program was implemented. However, in order to conclude
that the program caused this shift, it would be necessary to demonstrate that the change in sentencing
outcomes was not caused by shifts in the defendant/case characteristics that influence sentencing
decisions, such as age, gender, or prior criminal history. Using the proposed methodology with a
pre-post design (i.e., select a pre-program cohort and a post-program cohort), the analyses could
(after statistically controlling for any other changes between the pre- and post-period in the
characteristics of defendant that may influence sentencing) determine the degree to which the
probability of a prison sentence changed as a result of the implementation of the program. Similarly,
these data could be used to evaluate the degree to which changes in state sentencing policy influence
the likelihood that those convicted (or arrested/charged) of specific felony offenses are sentenced to
prison, such as the movement of a crime up (or down) in terms of its felony class, after accounting
for the other factors described here that could explain the imposition of a prison sentence.
Finally, given the lack of detailed, published, and publicly available statewide information on the
sentences imposed, the CHRI data should be analyzed to determine sentencing patterns on an annual
basis and the findings disseminated. At a minimum, information regarding the percent of those
convicted of a felony sentenced to prison and other forms of correctional supervision by age, race,
gender, and specific crime categories or felony classes should be published annually. Ideally, more
sophisticated analyses, such as those presented in this report, should be performed and published to
identify the degree to which specific defendant and current charge characteristics influence the
imposition of a prison sentence, and how those may have changed from previous years. Given how
current and up-to-date these data are (almost real-time), these types of analyses would provide policy
makers with timely information upon which to act and support data-drive policy making and
practice.
Appendix I
Table 3
Percent of Defendants Sentenced to Prison for a Felony Offense, by Defendant and Crime
Characteristics, Illinois 2012-2014 (N=115,434)
15 Under Illinois law, all individuals convicted of First Degree Murder and a Class X felony must be sentenced to prison. However, in a substantial portion of cases examined, individuals who were convicted of these offenses did not have any prison sentence recorded for the sentence imposed. Of those Murder and Class X felony cases not sentenced to prison, almost all received a sentence of “credit for time served” but without any specification that a prison sentence had also been imposed. There are a couple possible explanations to this pattern. First is that the prison sentence imposed was not recorded and only the portion of the sentence where the individual received credit for time served was recorded and posted to the individual’s criminal history record. It is also possible that while the original arrest may have been for Murder or a Class X felony, the final conviction offense was not, but that this change was not reflected in the CHRI data. Because we did not have any identifiers for the research subjects, we were not able to attempt to match the data to admission records by the Illinois Department of Corrections to determine if in fact a prison sentence had been imposed. However, these cases accounted for such a small percent of the overall sample, this issue did not impact the overall results or findings.
Category No Prison Sentence Prison Sentence Total
Total Sample
Average Age (in Years) (F=256,
p<.001)
31.9 32.7 32.3
Age
X2 = 198, df = 1, Phi = .04, p<.001
Under 25 54.8% 45.2% 100% 34.0%
25 and Older 50.4% 49.6% 100% 66.0%
Total 51.9% 51.9% 100% 100%
Race
X2 = 3,845, df = 1, Phi = .18, p<.001
White 61.1% 38.9% 100% 48.8%
Black 42.7% 57.3% 100% 51.2%
Total 51.7% 48.3% 100% 98.6%
Gender
X2 = 3,203, df = 1, Phi = .17, p<.001
Female 70.8% 29.2% 100% 16.2%
Male 48.2% 51.8% 100% 83.8%
Total 51.9% 48.1% 100% 100%
Conviction Offense Class
X2 = 5,742, df = 6, Cramer’s V = .22, p<.001
Murder15 18.1% 81.9% 100% 0.2%
Class X 21.0% 79.0% 100% 5.2%
Class 1 37.8% 62.2% 100% 10.5%
Class 2 45.9% 54.1% 100% 20.6%
Class 3 59.7% 40.3% 100% 19.3%
Class 4 57.8% 42.2% 100% 43.4%
Felony (Non-Specified) 89.5% 10.5% 100% 0.8%
Total 51.9% 48.1% 100% 100%
Conviction Crime Type
X2 = 2,653, df = 5, Cramer’s V = .15, p<.001
Violent 44.7% 55.3% 100% 12.0%
Weapons 25.9% 74.1% 100% 6.1%
Property 52.9% 47.1% 100% 28.3%
Drugs 55.2% 44.8% 100% 32.8%
Other 57.5% 42.5% 100% 18.6%
Missing 53.2% 46.8% 100% 2.2%
Total 51.9% 48.1% 100% 100%
At Least One Conviction Charge was Non-Probationable
X2 = 4,806, df = 1, Phi = .20, p<.001
None 55.3% 44.7% 100% 90.0%
One or More 21.3% 78.7% 100% 10.0%
UTotal 51.9% 48.1% 100% 100%
Defendant Received Credit for Time Served
X2 = 8,972, df = 1, Phi = .28, p<.001
No 68.6% 31.4% 100.0% 41.1%
Yes 40.2% 59.8% 100.0% 58.9%
Total 51.9% 48.1% 100% 100%
Arrest event involved some type of weapon charge
X2 = 2,032, df = 1, Phi = .13, p<.001
No 53.7% 46.3% 100.0% 92.5%
Yes 28.5% 71.5% 100.0% 7.5%
Total 51.9% 48.1% 100% 100%
Change in Number of Charges from Arrest to Sentencing
X2 = 12, df = 1, Phi = .01, p<.001
No change in number of charges
from arrest to sentencing
52.3% 47.7% 100.0% 58.6%
Number of charges reduced from
arrest to sentencing
51.3% 48.7% 100.0% 41.4%
Total 51.9% 48.1% 100% 100%
Change in Severity of Charges from Arrest to Sentencing
X2 = 14, df = 2, Cramer’s V = .01, p<.001
Severity of charges reduced from
arrest to sentencing
51.1% 48.9% 100.0% 8.3%
No change in severity of charges
from arrest to sentencing
52.0% 48.0% 100.0% 91.0%
Severity of charges increased from
arrest to sentencing
45.9% 54.1% 100.0% 0.7%
Total 51.9% 48.1% 100% 100%
Average Days from Arrest to
Case Disposition (F=96, p<001)
210 231 220
Prior Felony Conviction
X2 = 14,593, df = 1, Phi = .36, p<.001
None 71.9% 28.1% 100% 44.0%
One or More 36.1% 63.9% 100% 56.0%
Total 51.9% 48.1% 100% 100%
Prior Prison Sentence
X2 = 18,471, df = 1, Phi = .40, p<.001
None 66.9% 33.1% 100% 64.0%
One or More 25.2% 74.8% 100% 36.0%
Total 51.9% 48.1% 100% 100%
Prior Arrests
X2 = 15,662, df = 4, Cramer’s V = .37, p<.001
No Prior Arrests 78.7% 21.3% 100% 11.2%
1 to 4 Prior Arrests 69.7% 30.3% 100% 27.0%
5 to 8 Prior Arrests 54.1% 45.9% 100% 17.5%
9 to 12 Prior Arrests 43.8% 56.2% 100% 12.2%
13 or More Prior Arrests 29.3% 70.7% 100% 32.1%
Total 51.9% 48.1% 100% 100%
Prior Arrests X2 = 11,994, df = 1, Phi = .32, p<.001
Less than 5 72.3% 27.7% 100.0% 38.2%
5 or more 39.2% 60.8% 100.0% 61.8%
Total 51.9% 48.1% 100% 100%
Illinois Criminal Justice Information Authority300 W. Adams Street, Suite 200
Chicago, Illinois 60606Phone: 312.793.8408
Fax: 312.793.8422TDD: 312.793.4170
www.icjia.state.il.us
